Nucleic acid aptamers include DNA or RNA sequences that can recognize and specifically bind, often with high affinity, a particular molecule or ligand. See, for example, reports describing in vitro aptamer selection by Tuerk and Gold (1990) Science, 249:505-510, Ellington and Szostak (1990) Nature, 346:818-822, and Ellington and Szostak (1992) Nature, 355:850-852, as well as Jenison et al. (1994) Science, 263:1425-1429, which demonstrated the ability of an RNA aptamer to distinguish between theophylline and caffeine (which differ by a single methyl group) by four orders of magnitude. Similar to antibodies that bind specific antigens or receptors that bind specific molecules, aptamers are useful alone, to bind to a specific ligand (see, for example, Shi et al. (1999) Proc. Natl. Acad. Sci. USA, 96:10033-10038, which describes a multivalent RNA aptamer effective as a protein antagonist), and in combination, e.g., as a molecular “escort” for delivery of an agent to a specific location, cell, or tissue (see, for example, Hicke and Stephens (2000) J. Clin. Investigation, 106:923-928) or as part of a riboswitch. Riboswitches are complex folded RNA sequences including an aptamer domain for a specific ligand. Naturally occurring riboswitches have been found mainly in bacteria, and more recently in fungi (Kubodera et al. (2003) FEBS Lett., 555:516-520) and plants (Sudarsan et al. (2003) RNA, 9:644-647, which is incorporated by reference). Many riboswitches contain conserved domains within species (Barrick et al., (2004) Proc. Natl. Acad. Sci. USA, 101:6421-6426, which is incorporated by reference). Riboswitches that act in a “cis” fashion (i.e., that control expression of an operably linked sequence) are known to occur in the non-coding regions of mRNAs in prokaryotes, where they control gene expression by harnessing allosteric structural changes caused by ligand binding. For a review of “cis” riboswitches, see Mandal and Breaker (2004a) Nature Rev. Mol. Cell Biol., 5:451-463, which is incorporated by reference. Riboswitches that act in a “trans” fashion (i.e., that control expression of a sequence not operably linked to the riboswitch) have also been designed, see, for example, Bayer and Smolke (2005) Nature Biotechnol., 23:337-343, which is incorporated by reference.
Most known naturally occurring riboswitches are “off” switches, wherein the default state is “on” (i.e., the gene under the riboswitch's control is expressed), and ligand binding turns the gene “off”. In prokaryotes, these riboswitches have been found mainly in the 5′ untranslated region (5′ UTR) of mRNAs encoding biosynthesis genes; in eukaryotes, riboswitches have been found in the 3′ untranslated region (3′ UTR) or within introns (Sudarsan et al. (2003) RNA, 9:644-647; Templeton and Moorhead (2004) Plant Cell, 16:2252-2257). When an increased concentration of a particular metabolite or ligand is “sensed” by the riboswitch (bound by the aptamer domain), the riboswitch “switches off” gene expression through transcription termination and/or translation attenuation; see, for example, FIG. 2 in Mandal and Breaker (2004a) Nature Rev. Mol. Cell Biol., 5:451-463 and FIG. 4 in Sudarsan et al. (2003) RNA, 9:644-647.
At least two types of “on” riboswitches have been reported, wherein the default state is “off” and ligand binding turns the gene “on”. Expression of ydhL, encoding a purine exporter, is turned on by adenine binding to the ydhL aptamer; see Mandal and Breaker (2004b) Nature Struct. Mol. Biol., 11:29-35). Similarly lysine “on” riboswitches have been proposed to activate the expression of lysine exporter or degradation genes; see Rodionov et al. (2003) Nucleic Acids Res., 31:6748-6757. There are also lysine “off” riboswitches that control the expression of lysine biosynthesis genes; see Sudarsan et al. (2003) Genes Dev., 17:2688-2697.
A typical riboswitch is composed of an aptamer domain that remains largely conserved, and a regulatory domain that can vary more widely during evolution. In a non-limiting example, the coenzyme-B12 riboswitch controls gene expression by two main mechanisms, as dictated by the architecture of the regulatory domain (see FIG. 2 in Mandal and Breaker (2004a) Nature Rev. Mol. Cell Biol., 5:451-463). If the regulatory domain contains a “terminator stem”, the binding of coenzyme-B12 to its aptamer triggers transcriptional termination. If the expression platform contains an “anti-ribosome binding site stem”, the binding of coenzyme-B12 to its aptamer triggers translational attenuation. In some instances, it is believed that transcription and translation can be controlled simultaneously.
The present invention provides a novel transgenic plant having in its genome recombinant DNA that transcribes to at least one RNA aptamer to which a ligand binds, and can further include at least one regulatory RNA domain capable of regulating the target sequence. Depending on the design of the recombinant DNA, the regulatory RNA can act “in trans” or “in cis” in the transgenic plants to control expression of an endogenous or of an exogenous target sequence, and the ligand can be exogenous or endogenous. Transgenic plants of the invention are preferably stably transgenic plants in which a desired trait, or an altered trait, is achieved in the transgenic plant (or in a seed or progeny plant of the transgenic plant) according to whether or not the ligand is bound to the aptamer and the resulting expression (or suppression) of the target sequence.
Current methods to suppress a gene include, for example, the use of antisense, co-suppression, and RNA interference. Anti-sense gene suppression in plants is described by Shewmaker et al. in U.S. Pat. Nos. 5,107,065, 5453,566, and 5,759,829. Gene suppression in bacteria using DNA which is complementary to mRNA encoding the gene to be suppressed is disclosed by Inouye et al. in U.S. Pat. Nos. 5,190,931, 5,208,149, and 5,272,065. RNA interference or double-stranded RNA-mediated gene suppression has been described by, e.g., Redenbaugh et al. in “Safety Assessment of Genetically Engineered Fruits and Vegetables”, CRC Press, 1992; Chuang et al. (2000) PNAS, 97:4985-4990; Wesley et al. (2001) Plant J., 27:581-590.
The efficiency of anti-sense gene suppression is typically low. Redenbaugh et al. in “Safety Assessment of Genetically Engineered Fruits and Vegetables”, CRC Press, 1992, report a transformation efficiency ranging from 1% to 20% (page 113) for tomato transformed with a construct designed for anti-sense suppression of the polygalacturonase gene. Chuang et al. reported in PNAS, (2000) 97:4985-4990 that anti-sense constructs, sense constructs, and constructs where anti-sense and sense DNA are driven by separate promoters had either no, or weak, genetic interference effects as compared to potent and specific genetic interference effects from dsRNA constructs (see FIG. 1 and Table 1, PNAS, (2000) 97:4985-4990). See also Wesley et al. who report in The Plant Journal, (2001) 27:581-590, e.g., at Table 1, the comparative efficiency of hairpin RNA, sense constructs, and anti-sense constructs at silencing a range of genes in a range of plant species with a clear indication that the efficiency for anti-sense constructs is typically about an order of magnitude lower than the efficiency for hairpin RNA.
Matzke et al. in Chapter 3 (“Regulation of the Genome by double-stranded RNA”) of “RNAi—A Guide to Gene Silencing”, edited by Hannon, Cold Spring Harbor Laboratory Press, 2003, discuss the use of polyadenylation signals in promoter inverted repeat constructs. At page 58, they state that “the issue of whether to put polyadenylation signals in promoter inverted repeat constructs is unsettled because the nature of the RNA triggering RdDM [RNA-directed DNA methylation] is unresolved. Depending on whether short RNA or dsRNA is involved in RdDM, the decision to include a polyadenylation site might differ depending on the experimental system used. If dsRNA is involved in RdDM, then a polyadenylation signal is not required because dsRNA forms rapidly by intramolecular folding when the entire inverted repeat is transcribed. Indeed, nonpolyadenylated dsRNAs might be retained in the nucleus and induce RdDM more efficiently than polyadenylated dsRNAs. Matzke et al. continue: “If short RNAs guide homologous DNA methylation, then the situation in plants and mammals differ. In plants, which probably possess a nuclear form of Dicer, non-polyadenylated dsRNAs would still be optimal because they should feed preferentially into a nuclear pathway for dsRNA processing.”
Carmichael et al. in U.S. Pat. Nos. 5,908,779 and 6,265,167 disclose methods and constructs for expressing and accumulating anti-sense RNA in the nucleus using a construct that comprises a promoter, anti-sense sequences, and sequences encoding a cis- or trans-ribozyme. The cis-ribozyme is incorporated into the anti-sense construct in order to generate 3′ ends independently of the polyadenylation machinery and thereby inhibit transport of the RNA molecule to the cytoplasm. Carmichael demonstrated the use of the construct in mouse NIH 3T3 cells.
Various other nucleic acid constructs and methods for gene suppression have been described in recent publications. Shewmaker et al. (U.S. Pat. No. 5,107,565) disclose constructs for gene silencing that can contain two or more repetitive anti-sense sequence in tandem for modulating one or more genes. Resistance to a virus was achieved in a transgenic plant by use of a transgene containing a direct repeat of the virus's movement protein (Sijen et al. (1996) Plant Cell, 8:2277-2294). Another report demonstrated that nucleic acid constructs containing a promoter, a terminator, and direct or interrupted tandem repeats of either sense or anti-sense sequences, could induce gene silencing in plants (Ma and Mitra (2002) Plant J., 31:37-49. The expression of 1-aminocyclopropane-1-carboxylic acid (ACC) oxidase was downregulated in transgenic tomatoes containing a nucleic acid construct including a direct repeat of the ACC oxidase 5′ untranslated region sequence in the anti-sense orientation (Hamilton et al. (1998) Plant J., 15:737-346). Waterhouse and Wang (U.S. Patent Application Publication 2003/0165894) disclose a method for reducing phenotypic expression using nucleic acid constructs that transcribe to aberrant RNAs including unpolyadenylated RNAs. Clemente et al. disclose nucleic acid constructs including sense or anti-sense sequences lacking a normal 3′ untranslated region and optionally including a ribozyme, that transcribe to unpolyadenylated RNA. All of the patents cited in this paragraph are incorporated by reference in their entirety herein.
DNA is either coding (protein-coding) DNA or non-coding DNA. Non-coding DNA includes many kinds of non-translatable (non-protein-coding) sequence, including 5′ untranslated regions, promoters, enhancers, or other non-coding transcriptional regions, 3′ untranslated regions, terminators, and introns. The term “intron” is generally applied to segments of DNA (or the RNA transcribed from such segments) that are located between exons (protein-encoding segments of the DNA), wherein, during maturation of the messenger RNA, the introns present are enzymatically “spliced out” or removed from the RNA strand by a cleavage/ligation process that occurs in the nucleus in eukaryotes. Lin et al. (2003) Biochem. Biophys. Res. Comm., 310:754-760, and Lin et al. U.S. Patent Application Publications 2004/0106566 and 2004/0253604, which are incorporated by reference in their entirety herein, disclose methods for inducing gene silencing using nucleic acid constructs containing a gene silencing molecule (sense or anti-sense or both) within an intron flanked by multiple protein-coding exons, wherein, upon splicing and removal of the intron, the protein-coding exons are linked to form a mature mRNA encoding a protein with desired function and the gene silencing molecule is released.
However, apart from introns found between protein-encoding exons, there are other non-coding DNA sequences that can be spliced out of a maturing messenger RNA. One example of these are spliceable sequences that that have the ability to enhance expression in plants (in some cases, especially in monocots) of the downstream coding sequence; these spliceable sequences are naturally located in the 5′ untranslated region of some plant genes, as well as in some viral genes (e.g., the tobacco mosaic virus 5′ leader sequence or “omega” leader described as enhancing expression in plant genes by Gallie and Walbot (1992) Nucleic Acids Res., 20:4631-4638). These spliceable sequences or “expression-enhancing introns” can be artificially inserted in the 5′ untranslated region of a plant gene between the promoter but before any protein-coding exons. For example, it was reported that inserting a maize alcohol dehydrogenase (Zm-Adh1) or Bronze-1 expression-enhancing intron 3′ to a promoter (e.g., Adh1, cauliflower mosaic virus 35S, or nopaline synthase promoters) but 5′ to a protein-coding sequence (e.g., chloramphenicol acetyltransferase, luciferase, or neomycin phosphotransferase II) greatly stimulated expression of the protein (Callis et al. (1987) Genes Dev., 1:1183-1200). The Adh1 intron greatly stimulated expression of a reporter gene (Mascarenkas et al. (1990) Plant Mol. Biol., 15:913-920). Cis-acting elements that increase transcription of a downstream coding sequence in transformed plant cells were reported to occur in the 5′ untranslated region of the rice actin 1 (Os-Act1) gene (Wang et al. (1992) Mol. Cell Biol., 12:3399-3406). The rice Act1 gene was further characterized to contain a 5′ expression-enhancing intron that is located upstream of the first protein-coding exon and that is essential for efficient expression of coding sequence under the control of the Act1 promoter (McElroy et al. (1990) Plant Cell, 2:163-171). The Shrunken-1 (Sh-1) intron was reported to give about 10 times higher expression than constructs containing the Adh-1 intron (Vasil et al. (1989) Plant Physiol., 91:1575-1579). The maize sucrose synthase intron, when placed between a promoter and the first protein-coding exon, also increases expression of the encoded protein, and splicing of the intron is required for this enhanced expression to occur (Clancy and Hannah (2002) Plant Physiol., 130:918-929). Expression-enhancing introns have also been characterized for heat shock protein 18 (hsp18) (Silva et al. (1987) J. Cell Biol., 105:245) and the 82 kilodalton heat shock protein (hsp82) (Semrau et al. (1989) J. Cell Biol., 109, p. 39A, and Mettler et al. (May 1990) N.A.T.O. Advanced Studies Institute on Molecular Biology, Elmer, Bavaria). U.S. Pat. Nos. 5,593,874 and 5,859,347 describe improved recombinant plant genes including a chimeric plant gene with an expression-enhancing intron derived from the 70 kilodalton maize heat shock protein (hsp70) in the non-translated leader positioned 3′ from the gene promoter and 5′ from the first protein-coding exon. All of the patents and publications cited in this paragraph are incorporated by reference herein.
The present inventors have found that, unexpectedly, introns can be utilized to deliver a gene suppression element in the absence of any protein-coding exons (coding sequence). In the present invention, an intron, such as an expression-enhancing intron (preferred in certain embodiments), is interrupted by embedding within the intron a gene suppression element, wherein, upon transcription, the gene suppression element is excised from the intron to function in suppressing a target gene. Thus, no protein-coding exons are required to provide the gene suppressing function of the recombinant DNA constructs disclosed herein.
MicroRNAs (miRNAs) are non-protein coding RNAs, generally of between about 19 to about 25 nucleotides (commonly about 20-24 nucleotides in plants), that guide cleavage in trans of target transcripts, negatively regulating the expression of genes involved in various regulation and development pathways (Bartel (2004) Cell, 116:281-297). In some cases, miRNAs serve to guide in-phase processing of siRNA primary transcripts (see Allen et al. (2005) Cell, 121:207-221, which is incorporated herein by reference).
Some microRNA genes (MIR genes) have been identified and made publicly available in a database (“miRBase”, available on line at microrna.sanger.ac.uk/sequences). Additional MIR genes and mature miRNAs are also described in U.S. Patent Application Publications 2005/0120415 and 2005/144669A1, which is incorporated by reference herein. MIR genes have been reported to occur in inter-genic regions, both isolated and in clusters in the genome, but can also be located entirely or partially within introns of other genes (both protein-coding and non-protein-coding). For a recent review of miRNA biogenesis, see Kim (2005) Nature Rev. Mol. Cell Biol., 6:376-385. Transcription of MIR genes can be, at least in some cases, under promotional control of a MIR gene's own promoter. MIR gene transcription is probably generally mediated by RNA polymerase II (see, e.g., Aukerman. and Sakai (2003) Plant Cell, 15:2730-2741; Parizotto et al. (2004) Genes Dev., 18:2237-2242), and therefore could be amenable to gene silencing approaches that have been used in other polymerase II-transcribed genes. The primary transcript (which can be polycistronic) termed a “pri-miRNA”, a miRNA precursor molecule that can be quite large (several kilobases) and contains one or more local double-stranded or “hairpin” regions as well as the usual 5′ “cap” and polyadenylated tail of an mRNA. See, for example, FIG. 1 in Kim (2005) Nature Rev. Mol. Cell Biol., 6:376-385.
In animal cells, this pri-miRNA is believed to be “cropped” by the nuclear RNase III Drosha to produce a shorter miRNA precursor molecule known as a “pre-miRNA”. Following nuclear processing by Drosha, pre-miRNAs are exported to the nucleus where the enzyme Dicer generates the short, mature miRNAs. See, for example, Lee et al. (2002) EMBO Journal, 21:4663-4670; Reinhart et al. (2002) Genes & Dev., 16:161611626; Lund et al. (2004) Science, 303:95-98; and Millar and Waterhouse (2005) Funct. Integr Genomics, 5:129-135, which are incorporated by reference herein. In contrast, in plant cells, microRNA precursor molecules are believed to be largely processed in the nucleus. Whereas in animals both miRNAs and siRNAs are believed to result from activity of the same DICER enzyme, in plants miRNAs and siRNAs are formed by distinct DICER-like (DCL) enzymes, and in Arabidopsis a nuclear DCL enzyme is believed to be required for mature miRNA formation (Xie et al. (2004) PLoS Biol., 2:642-652, which is incorporated by reference herein). Additional reviews on microRNA biogenesis and function are found, for example, in Bartel (2004) Cell, 116:281-297; Murchison and Hannon (2004) Curr. Opin. Cell Biol., 16:223-229; and Dugas and Bartel (2004) Curr. Opin. Plant Biol., 7:512-520. MicroRNAs can thus be described in terms of RNA (e.g., RNA sequence of a mature miRNA or a miRNA precursor RNA molecule), or in terms of DNA (e.g., DNA sequence corresponding to a mature miRNA RNA sequence or DNA sequence encoding a MIR gene or fragment of a MIR gene or a miRNA precursor).
MIR gene families appear to be substantial, estimated to account for 1% of at least some genomes and capable of influencing or regulating expression of about a third of all genes (see, for example, Tomari et al. (2005) Curr. Biol., 15:R61-64; G. Tang (2005) Trends Biochem. Sci., 30:106-14; Kim (2005) Nature Rev. Mol. Cell Biol., 6:376-385). Because miRNAs are important regulatory elements in eukaryotes, including animals and plants, transgenic suppression of miRNAs could, for example, lead to the understanding of important biological processes or allow the manipulation of certain pathways useful, for example, in biotechnological applications. For example, miRNAs are involved in regulation of cellular differentiation, proliferation and apoptosis, and are probably involved in the pathology of at least some diseases, including cancer, where miRNAs may function variously as oncogenes or as tumor suppressors. See, for example, O'Donnell et al. (2005) Nature, 435:839-843; Cai et al. (2005) Proc. Natl. Acad. Sci. USA, 102:5570-5575; Morris and McManus (2005) Sci. STKE, pe41 (available online at stke.sciencemag.org/cgi/reprint/sigtrans;2005/297/pe41.pdf). MicroRNA (MIR) genes have identifying characteristics, including conservation among plant species, a stable foldback structure, and processing of a specific miRNA/miRNA* duplex by Dicer-like enzymes (Ambros et al. (2003) RNA, 9:277-279). These characteristics have been used to identify miRNAs and their corresponding genes in plants (Xie et al. (2005) Plant Physiol., 138:2145-2154; Jones-Rhoades and Bartel (2004) Mol. Cell, 14:787-799; Reinhart et al. (2002) Genes Dev., 16:1616-1626; Sunkar and Zhu (2004) Plant Cell, 16:2001-2019). Publicly available microRNA genes are catalogued at miRBase (Griffiths-Jones et al. (2003) Nucleic Acids Res., 31:439-441).
MiRNAs have been found to be expressed in very specific cell types in Arabidopsis (see, for example, Kidner and Martienssen (2004) Nature, 428:81-84, Millar and Gubler (2005) Plant Cell, 17:705-721). Suppression can be limited to a side, edge, or other division between cell types, and is believed to be required for proper cell type patterning and specification (see, for example, Palatnik et al. (2003) Nature, 425:257-263). Suppression of a GFP reporter gene containing an endogenous miR171 recognition site was found to limit expression to specific cells in transgenic Arabidopsis (Parizotto et al. (2004) Genes Dev., 18:2237-2242). Recognition sites of miRNAs have been validated in all regions of an mRNA, including the 5′ untranslated region, coding region, and 3′ untranslated region, indicating that the position of the miRNA target site relative to the coding sequence may not necessarily affect suppression (see, for example, Jones-Rhoades and Bartel (2004). Mol. Cell, 14:787-799, Rhoades et al. (2002) Cell, 110:513-520, Allen et al. (2004) Nat. Genet., 36:1282-1290, Sunkar and Zhu (2004) Plant Cell, 16:2001-2019).
The invention provides novel recombinant DNA constructs and methods for use thereof for suppression of production of mature miRNA in a cell, where the constructs are designed to target at least one miRNA precursor or at least one promoter of a miRNA precursor. Using constructs of the invention, suppression of production of mature miRNA can occur in the nucleus or in the cytoplasm or in both. In plants, microRNA precursor molecules are believed to be largely processed in the nucleus. Thus, in many preferred embodiments of the recombinant DNA construct of the invention, particularly (but not limited to) embodiments where the suppression occurs in a plant cell, suppression preferably occurs wholly or substantially in the nucleus. Another potential advantage of the invention is that miRNA precursors (especially pri-miRNAs, and to a lesser extent pre-miRNAs) offer substantially larger target sequences than does a mature miRNA.
In a preferred embodiment, the constructs and methods of the invention are designed to target nuclear-localized miRNA precursors (such as pri-miRNAs and pre-miRNA) prior to their export from the nucleus; such embodiments provide an advantage over conventional gene suppression constructs (e.g., containing inverted repeats) that typically result in accumulation of dsRNA in the cytoplasm. In such embodiments, recombinant DNA constructs of the invention include a gene suppression element designed to remain in the nucleus after transcription, for example, a gene suppression element that is transcribed to RNA lacking functional nuclear export signals. Such embodiments are particularly preferred for use, e.g., in plants, where processing of miRNA is believed to occur largely in the nucleus. In one preferred embodiment of the invention, the recombinant DNA construct includes a suppression element (e.g., one or more inverted repeats, anti-sense sequence, tandem repeats, or other suppression elements) embedded within a spliceable intron. The resulting suppression transcript remains in the nucleus, preferably resulting in the nuclear degradation of the target pri-miRNA or pre-miRNA, or alternatively, resulting in transcriptional silencing of a target MIR gene promoter, which, in turn, reduces the accumulation of the mature miRNA.
In other embodiments, recombinant DNA constructs of the invention include a suppression element transcribable to RNA that is exported from the nucleus to the cytoplasm, where, for example, the transcribed and exported RNA targets a cytoplasmic pre-miRNA. Such embodiments are particularly useful where miRNA processing at least partly occurs in the cytoplasm, e.g., in animal cells. In such embodiments, the suppression element is preferably transcribed to RNA including functional nuclear export signals.
In multicellular eukaryotes, including plants, microRNAs (miRNAs) regulate endogenous genes by a post-transcriptional cleavage mechanism in a cell-type specific manner. The invention further provides a recombinant DNA construct, and methods for the use thereof, wherein the construct includes transcribable DNA that transcribes to RNA including (a) at least one exogenous miRNA recognition site recognizable by a mature miRNA expressed in a specific cell, and (b) target RNA to be suppressed in the specific cell, whereby said target RNA is expressed in cells other than said specific cell. These constructs are useful for suppressing expression of a target RNA in a specific cell of a multicellular eukaryote (but allowing expression in other cells), including transcribing in the multicellular eukaryote a recombinant DNA construct including a promoter operably linked to DNA that transcribes to RNA including: (a) at least one exogenous miRNA recognition site recognizable by a mature miRNA expressed in a specific cell, and (b) target RNA to be suppressed in the specific cell, wherein the mature miRNA guides cleavage of target RNA in the specific cell, whereby expression of the target RNA is suppressed in the specific cell relative to its expression in cells lacking expression of the mature miRNA.
The present invention further provides novel mature miRNA sequences and MIR gene sequences from crop plants, including maize and soybean. The mature miRNAs processed from these genes belong to canonical families conserved across distantly related plant species. These MIR genes and their encoded mature miRNAs are useful, e.g., for modifying developmental pathways, e.g., by affecting cell differentiation or morphogenesis (see, for example, Palatnik et al. (2003) Nature, 425:257-263; Mallory et al. (2004) Curr. Biol., 14:1035-1046), to serve as sequence sources for engineered (non-naturally occurring) miRNAs that are designed to target sequences other than the transcripts targeted by the naturally occurring miRNA sequence (see, for example, Parizotto et al. (2004) Genes Dev., 18:2237-2242, and U.S. Patent Application Publications 2004/3411A1, 2005/0120415, which are incorporated by reference herein), and to stabilize dsRNA. A MIR gene itself (or its native 5′ or 3′ untranslated regions, or its native promoter or other elements involved in its transcription) is useful as a target sequence for gene suppression (e.g., by methods of the present invention), where suppression of the miRNA encoded by the MIR gene is desired. Promoters of MIR genes can have very specific expression patterns (e.g., cell-specific, tissue-specific, or temporally specific), and thus are useful in recombinant constructs to induce such specific transcription of a DNA sequence to which they are operably linked.